Agreed. But I would argue that "total amount of light" matters far more than "light per area".The same amount of light per square centimeter gets through the lens, regardless of the camera or sensor.
F-number is the established paradigm of photographers; how most of us think and work day in and day out. This mode of thought was necessary for film, since the response curve was intrinsically tied to the exposure (no matter what the sensor size), and it continues to work just fine in any discussion of a raw digital sensor of a single size.Originally Posted by George Slusher
However, in a discussion of raw digital sensors of multiple sizes, the old paradigm is suboptimal and even misleading. Now that we have a linear capture medium, it doesn't matter what the intensity of light per area is: all that matters is the *total* amount of light.
In film, if you go from 35mm to MF, you can't reduce exposure to keep DOF the same, because the image would be underexposed; that's why many MF/LF shooters used tripods and slow shutter speeds. But that's not true of digital: the size of the sensor compensates perfectly for reduced exposure, so any larger sensor can get the same image as a smaller sensor.
If you think in terms of f-number only, then it may seem that larger sensors have less noise, thinner DOF, less diffraction, higher MTF, more weight, etc. But that's not true in all circumstances: only when f-number is kept the same.
For example, people often lament that there is no FF equivalent to the 17-55 f/2.8 IS. But there is, and it's even better! The only reason they didn't see it is because they were held back by their f-number-centric frame of mind. The 24-105 has lower light intensity per area (exposure) but it has so much larger area that it more than makes up for it and actually collects more total light and has less total noise. You actually have to stop down the 24-105 by a third stop to get images similar to the 17-55.
As another example, some think that the f/2 lenses on four thirds are a great advantage over the f/2.8 lenses on full frame: they think they can get faster shutter speeds. In fact, it only takes an f/4 lens on FF35 to get the same shutter speed, noise, DoF, and diffraction as an f/2 lens on Four Thirds.
It's not always even possible to keep f-number the same when going to a larger sensor. Most medium format lenses are only f/2.8 or slower, and actually have deeper DOF (and lower total light gathering power in low light) than the f/1.2 lenses on FF35.
A tiny 5x4mm digicam at f/2.8 has the same light intensity as a 56x41mm f/2.8 medium format digital back. But the larger camera gathers for more light in total, and when printed at the same size, it has far less noise, less diffraction, and thinner DOF. However, if we apply the crop factor to f-number: f/31 on the MFDB will result in the same total light, same noise, same diffration, and same DOF.
When comparing multiple format sizes, keeping f-number is same is not necessary, desirable, or even possible in all circumstances.
Crop factor explains equivalence between two different sized sensors. Multiply the focal length by the crop factor to get the same angle of view. But angle of view is not the only thing that may be compared between two camera systems.
Noise, DOF, diffraction, MTF, MP, weight, etc. can also be compared, and these comparisons should be done by applying the crop factor to the f-number. The effect is to keep the iris diameter the same. Then, no matter what the sensor size, noise, DOF, and diffraction remains the same.
For non-macro focus distances, I find that the crop factor does provide equivalent depth of field to a very close approximation, so I find it highly useful.
I've done the calculations from 1/3" sensors up to 6x17 cm sensors and found that crop factor precisely correlates with DOF. Have you found any examples where it does not? As Mark stated, 50mm f/2.8 has the same DOF on APS-C as 80mm f/4.5 on FF35.
Reply With Quote